A Pilot Framework and Gap Analysis Towards Developing a Fluvial Classification System in the Ross Sea Region Antarctica

PCAS 17 (2014/2015)

Supervised Project

A PILOT FRAMEWORK AND GAP ANALYSIS TOWARDS DEVELOPING A FLUVIAL CLASSIFICATION SYSTEM IN THE ROSS SEA REGION ANTARCTICA

Lorna Louise Thurston

Student ID: 12670481

Abstract:

An integrated literature review has been undertaken with regards to the hydrological regime and fluvial geomorphology of the Ross Sea Region, Antarctica. The findings have been applied to develop a pilot framework for a process-based classification system of channels, ponds and lakes, and to identify gaps in knowledge that need to be addressed in order for the classification system to be developed further. The intention of the process-based classification system is that, once developed, it will be applied as a tool to help understand fluvial response to climate change and an increasing human footprint in the Ross Sea Region. In this regard, it would contribute towards a contemporary project - Assessing the Sensitivity of Dry Valleys to Change. It may also be useful for other applications, such as ecological research, and applicable to other regions of Antarctica. Several gaps in research have been identified that need to be addressed in order to integrate knowledge of the hydrological regime and fluvial morphology and subsequently develop a process-based classification system. In no particular order, these gaps include knowledge of: the spatial distribution of channel morphologies; fluvial morphological behaviour under heavily transport- and supply- limited conditions; the formation and desiccation of ponds, and their associated impact on the land’s surface; the significance, timing and origin of hill-slope processes; whether the spatial variability of melt, and the proportion of this melt that eventuates as surface flows, drive fluvial morphologies, or whether other processes exert a greater control; and whether events that are not directly climate/melt-driven, including when a glacier flows into and displaces a lake, jökulaups (ice-dam floods), and basal meltwater drainage of wet-based glaciers, have a transient or evolutionary effect on fluvial morphology.

CONTENTS 1. INTRODUCTION ...... 3 1.1 BACKGROUND ...... 3 1.2 PROPOSAL ...... 4 1.3 REPORT STRUCTURE ...... 5 2. SITE DESCRIPTION: ROSS SEA REGION, ANTARCTICA ...... 5 3. ROSS SEA REGION HYDROLOGICAL SYSTEM OVERVIEW ...... 9 3.1 SIGNIFICANCE OF THE HYDROLOGICAL SYSTEM ...... 9 3.2 A SUMMARY OF THE HYDROLOGICAL SYSTEM ...... 9 4. ATMOSPHERE ...... 11 5. GLACIERS AND SNOW ...... 13 5.1 GLACIAL / INTERGLACIAL HYDROLOGICAL FLUCTUATIONS ...... 13 5.2 SHORT-TERM FLUCTUATIONS OF GLACIERS AND SNOW PATCHES ...... 14 5.3 ICE CLIFFS VS. GLACIERS’ SURFACES ...... 15 6. CRYOCONITE HOLES ...... 16 7. MELTWATER STREAMS ...... 17 8. THE ONYX RIVER ...... 18 9. LAKES ...... 20 9.1 DEFINING LAKES ...... 20 9.2 TAYLOR VALLEY LAKES ...... 20 9.3 INPUTS AND OUTPUTS ...... 21 9.4 DRIVERS OF LAKE LEVEL CHANGE ...... 21 10. PONDS ...... 22 11. SUBSURFACE FLOWS ...... 24 12. A PILOT FRAMEWORK FOR DEVELOPING A CLASSIFICATION SYSTEM ...... 25 12.1 STREAMS AND RIVERS ...... 25 12.2 PONDS AND CRYCONITE HOLES ...... 27 12.3 LAKES ...... 28 13. CONCLUSION ...... 28 14. REFERENCES ...... 29

1. INTRODUCTION 1.1 BACKGROUND Fluvial morphological classifications systematically identify water bodies, reaches or units with relatively homogenous morphologies in order to easily communicate information about their characteristics. They are based on the concept that fluvial geomorphology responds to changes in environmental forcing - natural or anthropogenic. Ultimately they seek to predict how a water body, reach or unit will respond over time to environmental forcing. They also provide a means for hydrological data to be extrapolated to different localities (Rosgen, 1994), although this requires considerable caution. Accordingly, such classifications assist in ecological, geological, engineering and resource management applications, investigations of channel response to climate change and human impact, and mass balance calculations.

The advent of assigning and applying classifications is not new, with literature citing references at least as far back as 1899 when Davis divided streams into youthful, mature, and old age according to their relative stage of adjustment (McDavitt, 2004; Rosgen, 1994). Since this time, geomorphologial literature has been directed towards channel classifications, rather than classifications of other water bodies, which is likely because of their dynamic nature. Several classification systems concerning water chemistry / ecological health have been developed to assist with managing lakes (Salm, Saros, & Fritz, 2009; Uttormark & Wall, 1975; Wagner, Bremigan, & Cheruvelil, 2007), but geomorphological classifications have not eventuated as they have for channels. This is attributed to lakes considerably slower rate of morphological change, comparative to channels, which are dynamic hydrological features.

Channel classification systems have progressed considerably since Davis’s simple classification. Horton (1945) paved the way for quantitative fluvial geomorphology, introducing the concept of stream ordering which was developed further by Strahler (1957). Under this system, first order streams emerge at the headwaters of a catchment and orders progressively increasing at each confluence. Broad-ranging descriptions of channel form were also published – starting with simple classifications such as straight, meandering and braided (Leopold and Wolman, 1957), and progressing to more detailed descriptions (Kellerhals et al., 1976).

Rosgen’s (1994) classification system, which establishes quantitative groupings of homogenous reaches in natural rivers, has been widely applied. It focuses on providing a broad geomorphic characterization (Level I) and morphological description (Level II) of channels. Eight generalized categories of “stream types” are identified using longitudinal profiles, valley and channel cross-sections, and plan-view patterns. Streams are further categorized into 42 types based on discreet slope ranges and dominant channel-material particle sizes.

More recently the focus has been on classification systems that couple morphology and process in specific environments. Montgomery and Buffington’s (1997, pp. 1) process-based system identifies seven distinct mountain reach types, including: “colluvial, bedrock, and five alluvial channel types: cascade, step pool, plane bed, pool riffle, and dune ripple”. This allows dominant channel forming mechanisms to be

identified based on fluvial morphology, which may be hillslope processes (supplying the system) or fluvial processes (transporting material in the system).

Although the hydrological regime of Antarctica has been investigated towards ecological ends (Fountain et al., 1999; Gooseff et al., 2011), a comprehensive classification system for fluvial systems has not been developed for polar environments, which presents a significant gap in knowledge. It is proposed that such a classification system would assist with the broad-ranging applications listed in the opening paragraph. As Antarctica is the last place on earth to feel human impact, and is sensitive to climate forcing (as described in section 2), it presents an ideal environment to investigate fluvial morphological responses to climate change. The increasing human footprint in Antarctica means that assessing fluvial and associated ecosystem response to anthropogenic forcing is also of contemporary interest. It is considered that developing a classification system for freshwater systems will assist with research towards this end and contribute towards a wider project – Assessing the Sensitivity of the Dry Valleys to Change.

Antarctic, Arctic and Icelandic channels often have unusual physical features that depart from the classifications applied in other natural environments. For example, freeze-thaw cycles, snow and channel ice, ice-dams, permafrost, peat soils and sparse vegetation may distinguish unique morphologies at the poles (Crawford and Stanley, 2014; Gore, 1992). Small climatic changes can lead to rapid and significant variations in hydrological regime, known as polar amplification (Foutain, Lyons, & Burkins et al., 1999; Gooseff, McKnight, & Doran et al., 2011). This is attributed at least partly to considerable ice reservoirs in polar regions, with fluctuations of temperature above and below 0° determining the presence or absence of liquid water (Fountain et al., 1999; Gooseff et al., 2011; Howard-Williams & Hawes, 2007).

However, Antarctic, Arctic and Icelandic glacio-hydrological regimes and associated morphologies will not necessarily be analogous owing to climatic differences (Smith, 2000). For this reason, it is proposed that a morphological classification system specific to the Ross Sea Region of Antarctic is developed, rather than a generic classification for polar environments. The Ross Sea Region drains approximately one- fourth of Antarctica (Denton et al., 1989) as described further in section 2 below. Hydrological research in the Ross Sea Region is skewed spatially towards the McMurdo Dry Valleys and Taylor Valley in particular. This is because liquid water driving hydrological processes principally exists in ice-free areas, the McMurdo Dry Valleys are the largest ice-free area in Antarctica, and because the National Science Foundation’s Long-term Ecological Research project has focused research in Taylor Valley (Levy et al., 2014).

1.2 PROPOSAL A research project is proposed to develop an inventory of information on hydrological processes and associated fluvial morphologies in the Ross Sea Region. This information will provide the framework for developing a classification system. The project has the following aims:  Identify the hydrological regime and fluvial morphological features of the Ross Sea Region of Antarctica.

 Consider the quality of information available on the hydrological regime and fluvial morphology in the Ross Sea Region, and determine whether (and what) additional information would be required in order to develop a classification system.

1.3 REPORT STRUCTURE In order to address the above aims, this report is divided into the following main sections:  Site Description;  Ross Sea Region Hydrological System Overview;  Atmosphere;  Glaciers and Snow;  Cryconite Holes;  Meltwater Streams;  The Onyx River;  Lakes;  Ponds;  Subsurface Flows;  Developing a Classification System; and  Conclusion.

2. SITE DESCRIPTION: ROSS SEA REGION, ANTARCTICA The Ross Sea Region is a geographical area at the margin of the West Antarctic and East Antarctic Ice Sheets. The Region encompasses part of the Antarctic continent, including part of the Transantarctic Mountains in the west, the largest ice shelf in the world – the Ross Ice Shelf, which extends from the continent, and part of the Southern Ocean and its islands (Clarkson, 2007). Its geographical boundaries are not strictly defined in literature, but it generally lie between 150°E and 150°W, and south of 60° Latitude (Figure 1).

The environment is characterised by extreme cold, extreme aridity, and katabatic winds, which generally blow from the polar plateau towards the coast. The Ross Sea is covered by ice throughout most of the year, although this occasionally breaks out during the austral summer (Vincent and James, 1996).

The Transantarctic Mountains, which separate East and West Antarctica, extend through the Region. The 3000km long and 750-1000km wide West Antarctic Rift System has resulted in a volcanic environment along the elevated topography of the Transantarctic Mountains and in Marie Byrd Land on the other side of the Ross Ice Shelf, owing to thinning of the lithosphere (Cooper, Adam, & Coulter et al., 2007). Active volcanism is currently occurring at Mount Erebus. Mass wasting processes also occur, including debris flows and debris avalanches (Levy et al., 2014).

The Ross Sea Region is estimated to drain one-fourth of Antarctica (Denton et al., 1989). The catchment has been estimated to encompass 1,650,000km2 of the East Antarctic Ice Sheet, on which ice flows via outlet glaciers in the Transantarctic Mountains into the Ross Ice Shelf or Ross Sea (Denton et al., 1989). The catchment encompasses approximately 750,000km2 of the West Antarctic Ice Sheet, on which ice flows via major ice streams into the Ross Ice Shelf (Denton et al., 1989).

The Ross Sea Region comprises mainly ice-covered areas, including the grounded Ross Ice Shelf, alpine glaciers and outlet glaciers to the sea. The McMurdo Ice Shelf, at approximately 78°S, 1500-2000km2 and 10-50m thick, is a small part of the Ross Ice Shelf (Figure 2) (Hawes, Safi, & Sorrell et al., 2011; Vincent and James, 1996). The nearest neighbours to the Ross Sea Region containing lakes and streams are the Bunger Hills and Vestfold Hills in East Antarctica, about 2400km from McMurdo Sound (refer to the inset map shown in Figure 1) (Vincent and James, 1996).

Ice-free areas occur within the McMurdo Dry Valleys and other ice-free niches within the Ross Sea Region, including Cape Crozier (18km2), Cape Bird (15km2), and Cape Royds (13km2) on Ross Island (Broady, 1989; Nie, Liu, & Sun et al., 2012), and the Kar Plateau in Southern Victoria Land (Seppelt, Green, & Schroeter, 1995). The southernmost aquatic ecosystems occur in the Darwin Glacier Region at 80°S latitude, approximately 300km south of the McMurdo Dry Valleys (Vincent and James, 1996). Please refer to Figure 2 for a map showing these localities. At 44,890km2 ice-free regions only occupy approximately 0.32% of the entire Antarctic continent (Clarkson, 2007), and over 15% of these ice-free areas occur within the McMurdo Dry Valleys within the Ross Sea Region (Colesie, Gommeaux, & Green, 2013).

The McMurdo Dry Valleys are of significant ecological interest owing to the delicate nature of its ecosystems that are dependent on hydrological processes generating liquid water (MacDonell, Fitzsimons, & Mölg, 2013). As aforementioned, a Long-term Ecological Research project, driven by the National Science Foundation, has led to much of the ecological and associated hydrological research in the Ross Sea Region being focused in the McMurdo Dry Valleys, in particularly Taylor Valley (Figure 2).

The McMurdo Dry Valleys are an advent of the Transantarctic Mountains blocking flow of the East Antarctic Ice Sheet toward McMurdo Sound, and ablation exceeding accumulation on the valley floors (Fountain et al., 1999). They comprise of the Taylor Valley, Wright Valley and Victoria Valley (south – north) in southern Victoria Land, which together occupy an area of approximately 15,000km2, 4000km2 of which is ice-

Figure 1: A map showing the Ross Sea Region of Antarctica, with lakes and streams emphasized. The map was produced using ArcGIS.

Figure 2: A map showing the areas of the Ross Sea Region referred to in-text where hydrological and fluvial morphology research is focused, with lakes and streams emphasized. The map was produced using ArcGIS.

free between approximately 76°30’-78°30’ South and 160°-164° East (Fountain et al., 1999; Green and Lyons, 2009). Glaciers occupy approximately 35% of Taylor Valley (Fountain et al., 2006). Marine and lacustrine sediments and associated organic matter, derived from increased sea levels during the Pliocene and lakes present since that time, are found in the Dry Valleys (Fountain et al., 1999). Fish and insects do not occur within the freshwater ecosystems, and the largest animals reported to exist in the extreme conditions of the McMurdo Dry Valleys are soil invertebrates (Doran, Priscu, & Lyons et al., 2002; Vincent and James, 1996).

The Ross Sea Region has a human presence, albeit limited given Antarctica was the last continent on earth to be subjected to human impacts. New Zealand has a territorial claim to the majority of the Ross Sea Region (150°W to 160°E), and Australia has a territorial claim to remaining slice (160°E to 150°E) (Waterhouse, 2001). These claims are currently “frozen” by the Antarctic Treaty (1959) – Article IV. Within the Ross Sea Region, New Zealand has established a permanent base - Scott Base (Figure 1). The United States does not have any territorial claims to Antarctica, but they have a permanent base within the Ross Sea Region – McMurdo Station, as do the Italians – Stazione Mario Zucchelli (Figure 1). These permanent bases, along with other temporary bases, campsites and access by vessels, allow for human presence pursuant to the Antarctic Treaty System.

3. ROSS SEA REGION HYDROLOGICAL SYSTEM OVERVIEW 3.1 SIGNIFICANCE OF THE HYDROLOGICAL SYSTEM Antarctica contains approximately 90% of global freshwater resources in the form of ice (Hulbe, 2007). The hydrology of the continent is poignant to human interests with regards to: its water resources, the role of the Antarctic continent in the global environment system, the consequences of ice melt, climate change research, and ecological research and associated benefits. Understanding hydrological processes is critical for understanding the ecology of the Antarctic continent as liquid water is a necessity of life, albeit terrestrial ecosystems ultimately depending on a range of soil physical and chemical factors (Fountain et al., 1999; Levy, Fountain, & Gooseff et al., 2014; Howard-Williams & Hawes, 2007). Understanding hydrological processes, coupled with an understanding of morphological response, is also important for assessing how the environment may support, and respond to, an increasing human footprint.

3.2 A SUMMARY OF THE HYDROLOGICAL SYSTEM Hydrological processes of the Ross Sea Region are largely controlled by climatic conditions that allow ice to liquefy, which vary in both space and time. Above 1500masl the majority of the Region is frozen, but below this altitude conditions are seasonally suitable for melt to occur during the austral summer under a positive surface energy balance1 (Gooseff et al., 2011; Howard-Williams & Hawes, 2007). Precipitation, predominantly in the form of snow, is also greatest in summer. In the shadow of the Transantarctic Mountains, precipitation is very low (< 10cm water equivalent per year) and exerts a secondary influence on the hydrological cycle

1 Surface energy balance is controlled by net radiation, sensible and latent heat exchanges. 9

comparative to surface energy balance1 (Gooseff et al., 2011; Howard-Williams & Hawes, 2007). Sublimation, evaporation and melt processes occur (Fountain et al., 1999). Wind exerts influence on both accumulation and ablation, dependent on spatial and event variability.

The landscape is barren and there is no vegetation buffer between atmospheric and land-surface hydrological exchanges. Limited biota also means that climate-driven hydrological processes exert a significant influence on the Region’s geomorphology (Fountain et al., 1999). Furthermore, complications in hydraulic processes and associated morphology associated with woody debris do not exist.

The ablation zone of ice shelves, glaciers, and snow cover are affected by sublimation and melt processes (Gooseff et al., 2011). Snow patches occur on valley bottoms and generally sublimate, rather than melt (Chinn, 1981; Fountain et al., 1999). However, wind-blown snow accumulated at glacier termini (Fountain et al., 1999), perennial snow cover on lakes, and snow collected in nivation hollows, contribute to spring melt, with the latter trickling down valley walls in streaks (Gooseff et al., 2011). Snow patch longevity is seasonally and locally variable, being greater in winter and affected by local wind events and topography (Gooseff et al., 2011).

Glaciers and ice shelves are critical sources of melt processes, feeding the hydrological system. Small changes in energy balance can lead to either melt or desiccation (Howard-Williams & Hawes, 2007). Meltwater generation from ice shelves is reportedly most extensive on the McMurdo Ice Shelf, where 60% of the 1500- 2000km2 area is covered by meltwater at the end of summer (Vincent and James, 1996). With regards to glaciers, melt is locally variable, dependent on local climatic conditions, and accounts for approximately 30% of ablation from glaciers (Fountain et al., 1999). A few centimetres of snow cover on glaciers can reportedly inhibit melt processes owing to increased albedo (Gooseff et al., 2011), whereas sediment cover increases albedo and leads to cryconite hole formation (MacDonnell, Fitzsimons, and Mölg, 2013).

Both wet-based and cold-based glaciers occur in the Ross Sea Region and in the McMurdo Dry Valleys, although literature does not report this consistently (e.g. Chinn, 1981; Fountain et al., 1999; Gooseff et al., 2011). Wet-based glaciers are influenced by thicker ice cover which enables pressure melting point to be reached and subglacial flows to occur (I. Hawes, personal communication, February 16, 2015). Alpine glaciers are abundant within the McMurdo Dry Valleys and some outlet glaciers occur, with seasonal melt occurring in the ablation zones.

Under the influence of seasonal melt, cryconite holes and ponds occur on the surfaces of glaciers, streams and rivers flow ephemerally, ponds form, or are replenished, in topographical depressions in ice-free areas, and the peripheries of lakes melt. Evaporation occurs from these surface water bodies, which may be in the order of 6mm/day (Gooseff et al., 2011).

Ephemeral flows in the austral summer replenish the lakes. Although solar radiation occurs for 24 hours each day in summer, temporal variations occur with climatic variations affecting energy balance, the position of the sun relative to ice faces, and

snowfall. Ice cliffs generate significant melt volumes, despite their small surface areas comparative to glaciers’ surfaces, especially during cooler periods (Lewis, Fountain, & Dana, 1999).

Although climate is the first-order driver of the climate signal, a catchment’s physiography variably modifies the climate signal (Laizé & Hannah, 2010). Flow events that are not directly “climatic” occur, including: when a glacier flows into and displaces a lake, jökulaups (ice-dam floods), and basal meltwater drainage of wet- based glaciers (Chinn, 1981). Such events can lead to replenishment outside the austral melt season. In addition, sediment sources for erosion, transportation and deposition processes exert influence on hydraulic processes and resulting fluvial morphology (Mosley et al., 1988).

Stream – soil connections occur through hyporheic zones, which are thawed areas adjacent to and under stream channels that visually contrast adjacent dry soils (Gooseff et al., 2011; Levy et al., 2014). Hyporheic zones are limiting factors in terms of stream flow, because they fill and become porous before water can flow downstream. The hyporheic zone may inhibit meltwater from reaching lakes during low-melt summers, especially in longer streams because of the significant volumetric storage that first needs to be satisfied prior to surface flows commencing (Gooseff et al., 2011).

Sub-surface flow is limited as the landscape is underlain by permafrost to depths greater than 100m and up to 600m. However, permafrost melt occurs in the austral summer resulting in approximately 10-70cm of unfrozen soils (Gooseff et al., 2011; Levy et al., 2014), and limited groundwater flow below this is reported to occur (Fountain et al., 1999). Ice streams and subglacial lakes occur below the ice surface, variably modifying ice flow (Carter, Blankenship, & Peters et al., 2007), but these subglacial processes are beyond the context of this investigation.

4. ATMOSPHERE The atmosphere is one of the hydrological reservoirs of the Ross Sea Region. It is conceptualized that large-scale atmospheric circulation influences regional climate, including precipitation and air temperature, which act as “inputs” into a catchment’s hydrological system. Inputs are modified by a catchment’s physiography and feedback mechanisms between the catchment and regional climate (Laizѐ and Hannah, 2010; Lavers, Prudhomme, & Hannah, 2010). Discharge and sediment yield act as outputs of the system and may be measured, analysed and interpreted with regards to the system’s inputs and changes over space and time.

Relatively little research into the relationship between large-scale atmospheric circulation and discharge has been undertaken in the southern hemisphere, comparative to the northern hemisphere. Research that has been undertaken highlights the significance of the El Nino Southern Oscillation (“ENSO”) and Southern Annular Mode (“SAM”), and suggests that discharge responds to El Ninõ and La Ninã events (Araneo & Compagnucci, 2008; Kiem & Verdon-Kidd, 2010; Poveda & Mesa, 1996).

At the regional scale, temperature is found to be significant in driving hydrological processes (Lewis, Fountain, & Dana, 1999; Gooseff et al., 2011). In the McMurdo Dry Valleys the average annual temperature is approximately -18° to -20°, with winter temperatures of -65°, and summer temperatures close to freezing (Doran, Lyons, & McKnight, 2002). However, McConchie & Hawke (2002) caution that correlating temperature and streamflow is overly simplistic, because the amount and intensity of incoming solar radiation is in fact the dominant control in Miers Valley (Figure 2). Streamflow and water temperature are both largely be a function of net shortwave radiation (Cozzetto, McKnight, & Nylen, 2006; McConchie & Hawke, 2002).

Katabatic winds are significant, especially with regards to local mass balance and temperature. They erode, transport, and deposit snow, redistributing it throughout the valleys and across ice shelves. Katabatic winds vary in space, increasing up-valley, reportedly by 14% for every 10km in winter (Nylen, Fountain, & Doran, 2004). They are affected by physiographic factors, which influence local environmental variations. For example, the western end of Taylor Valley at Lake Bonney is warmer and windier than the eastern end at Lake Fryxell (Fountain et al., 1999). Fountain et al. (1999) have attributed this local variation to two reasons: 1) the interaction between dry katabatic winds of the ice sheet and coastal breezes of the sea ice resulting in an up-valley (Lake Fryxell to Lake Bonney) increase in air temperature and wind speed; 2) the Nussbaum Riegel confines the effects of katabatic winds to the western and eastern parts of the valley and retards coastal storms from reaching Lake Bonney. Sublimation and associated ablation are consequently elevated at Lake Bonney compared with Lake Fryxell, demonstrating the significance of local climate on hydrological processes.

The frequency of katabatic winds exerts a positive influence on temperature, with frequencies calculated to increase average annual temperatures by 0.7°C to 2.2°C in the McMurdo Dry Valleys (Nylen, Fountain, & Doran, 2004). The frequency of katabatic wind events is elevated in winter. Temperature reportedly increases 1°C for every 1% increase in katabatic frequency in winter, and can therefore increase winter air temperatures by up to 30°C (Nylen, Fountain, & Doran, 2004).

Precipitation is also significant at the local scale, albeit not exerting the same regional impact as temperature and wind. Average precipitation is estimated to be 2-60mm water equivalent, predominately from snowfall (Gooseff et al., 2011; MacDonell, Fitzsimons, & Mölg, 2013), and decreasing westward away from the ocean (Fountain et al., 1999). A single storm can generate significant snowfall, given the low annual volume (Fountain et al., 2006). Spatial variation occurs in the same manner as for wind, with Taylor Valley becoming drier, warmer, and windier up-valley for example (Foutain et al., 2006).

Variations in snowfall between Lake Bonney and Lake Fryxell have been found to affect discharge. Annual discharge is consistent amongst glaciers within the catchment of Lake Bonney, which is attributed to the lack of snow (Fountain et al., 1999). Conversely, annual discharge is variable between glaciers within the catchment of Lake Fryxell because of the spatial variability of snowfall and its redistribution by wind (Fountain et al., 1999). Lake Bonney is rising faster than Lakes Fryxell and Hoare, which may be due to the lack of snow meaning that albedo is not reduced and associated melt is not hampered (Fountain et al., 1999).

5. GLACIERS AND SNOW 5.1 GLACIAL / INTERGLACIAL HYDROLOGICAL FLUCTUATIONS As in other areas of Antarctica, the Ross Sea Region is affected by long-term fluctuations of ice expansion and retreat driven by long-term climate oscillations (Anderson, Conway, & Bart et al., 2014; Denton et al., 1989). The Ross Ice Shelf has experienced periodic shifts in mass balance 2 , including periods of expanding, thickening and grounding, and periods of shrinking, thinning and floating. During periods of greater accumulation, ice-coverage spread into the seaward end, of otherwise ice-free, McMurdo Dry Valleys (Chinn, 1981).

Reconstructions indicate that an expanded grounded Ross Ice Sheet occupied the Ross Embayment during the Last Glacial Maximum (“LGM”)3 (Anderson et al., 2014), although reconstructions have previously suggested ice in the outer Embayment could have been floating (Denton et al., 1989). The expansion of the Ross Ice Shelf prior to this time caused lobes of sea ice to close off Taylor Valley, Wright Valley and Garwood Valley, resulting in the formation of glacial Lake Washburn (>100m deep) (Chinn, 1981; Fountain et al., 1999; Howard-Williams & Hawes, 2007; Kellogg, Stuvier, & Kellogg, 1980; Levy, Fountain, & O’Conner et al., 2013), Great Lake Vanda (>200m deep) (Howard-Williams & Hawes, 2007), and an expanded lake in the Garwood Valley (Levy et al., 2013). Lake Washburn occupied much of Taylor Valley and was some 340m above current lake levels (Chinn, 1981; Kellogg et al., 1980).

Studies have returned varying results with regards to the timing of massive late Wisconsin / Holocene retreat of grounded ice in the Ross Sea Region. Ice may have retreated in step with widespread massive retreat of West Antarctica coming out of the LGM, commencing at 13-15ka (Denton et al., 1989; Clark, Dyke and Shakun et al., 2009), or may have retreated from the outer continental shelf prior to the LGM (Anderson et al., 2014).

Howard-Williams & Hawes (2007) reported that lakes were at their maximum around 10-12ka, coming out of the Younger Dryas cooling event. As ice retreated, drainage of Lake Washburn ensued and lake levels fell dramatically until water was contained by topography (Kellogg et al., 1980; Howard-Williams & Hawes, 2007). Lakes Washburn and Vanda are thought to have drained by approximately 7ka (Fountain et al., 1999), with the drainage of the lake in Garwood Valley further south commencing around 7.3ka, and continuing through to 5.5ka, when the grounding line moved past the Valley’s mouth (Levy et al., 2013).

Ice in the Ross Sea Region is no longer retreating, resulting in a positive mass balance, close to equilibrium (Anderson et al., 2014; Fountain et al., 2006). This may reflect a reversal of the long-term Holocene retreat of the ice sheet, or a century-scale fluctuation of the ice streams (Anderson et al., 2014). Lake levels are still reportedly rising, with periods of stability (Howard-Williams & Hawes, 2007).

2 Mass balance asserts the difference between ice accumulation and ablation, and is applied to calculate and forecast climate change and sea level rise (Pritchard, Ligtenberg, & Fricker et al., 2012). 3 The LGM occurred between 26.5ka and 19ka (Clark, Dyke and Shakun et al., 2009). 13

5.2 SHORT-TERM FLUCTUATIONS OF GLACIERS AND SNOW PATCHES Glaciers occur in outlet and alpine locations and snow patches cover approximately 10% of valley floors in summer (Cozzetto et al., 2006). The majority of glaciers in the McMurdo Dry Valleys are alpine glaciers, with outlet glaciers including Taylor Glacier and Wright Upper Glacier that are termini of the East Antarctic Ice Sheet (Gooseff et al., 2011). At third glacier system – comprising of Ross Sea ice has influenced the region in the past as described above, but is not present today (Kellogg et al., 1980). Mass gains and losses from glaciers are reportedly small compared with temperate glaciers and they are relatively stable (Fountain et al., 1999; Fountain et al., 2006). Alpine glaciers have increased in length and reduced in width since 3100-5000 years BP when they were shorter and wider, but overall variations in mass balance and flow rates are reportedly minor (Chinn, 1981).

Glaciers may be considered with regards to spatial and temporal variability. At the local spatial scale, accumulation may be driven by system inputs (precipitation and wind deposition for example) increasing and/or system outputs (wind erosion and net evaporation for example) decreasing (Pritchard, Ligtenberg, & Fricker et al., 2012). Knuth, Tripli, & Thom et al. (2010) asserts that precipitation (P) and wind (Q) are the two critical factors influencing snow depth change on the Ross Ice Shelf, and distinguishing between these two forms of accumulation is significant to mass balance.4 However, avalanches and calving of ice cliffs also play a role (Fountain et al., 2006).

Seasonal fluctuations occur, with accumulation being greatest in summer and lesser in winter (Fountain et al., 2006). Warmer summer temperatures are associated with increased precipitation allowing snow and ice to accumulate, countering ablation, which is also elevated in summer. Colder temperatures are associated with miniscule

4 The factors influencing snow depth may be summarised by the following equation:

B = P - E - Q - R B = total snow depth change; P = accumulation by precipitation (i.e. snowfall from synoptic or mesoscale systems); E = snow loss from net evaporation (sublimation minus deposition); Q = snow loss or gain due to the horizontal flux of snow (i.e. wind); and R = loss of snow from meltwater.

The wind factor can be explained by the following equation:

Q = Qt – Qs – Qc Q = changes in depth due to wind; Qt = transport of snow particles by wind; Qs = net sublimation of the snow particle; and Qc = changes in snow by compaction.

(Knuth et al., 2010).

precipitation and also with reduced ablation. Hydrology is therefore largely driven by summer conditions.

Melt of glaciers and snow patches is driven by surface energy balance (net radiation, sensible and latent heat exchanges), with precipitation being of secondary importance (Gooseff et al., 2011). Sublimation, driven by wind, temperature, radiation, and atmospheric water content, is dominant from snow patches and clean ice surfaces (Gooseff et al., 2011). In contrast, melt is increased on glaciers with dirty surfaces, as discussed in relation to cryconite hole formation below (MacDonnell, Fitzsimons, and Mölg, 2013). Melt occurs below 1500m and supraglacial hydrological systems develop, which may include cryconite holes, ponds, and/or incised and/or near-surface channels (MacDonell, Fitzsimons, & Mölg, 2013). Snowfall can block flow in channels during the early part of the melt season (Chinn, 1981; Fountain et al., 1999).

5.3 ICE CLIFFS VS. GLACIERS’ SURFACES Melt occurs from both the surfaces’ of glaciers and from terminal cliff faces in the ablation zone. Ice cliffs have been found to generate significant meltwater volumes, despite their small area compared with glaciers’ surfaces, because of relatively constant albedo and hampered sublimation (discussed in the following paragraphs). For example, the terminus cliffs represent only 2% of the ice surface on the Canada Glacier in Taylor Valley, but have been found to account for 15-20% of meltwater runoff (Lewis, Fountain, & Dana, 1999).

Snowfall reduces albedo and can therefore act to limit or inhibit meltwater generation from glaciers’ surfaces (Gooseff et al., 2011). As ice cliffs are not subject to snowfall, albedo remains relatively constant, which means shortwave radiation drives ablation throughout the entire summer season. Incoming long-wave radiation is also elevated on cliff faces, increasing net radiation and driving melt (Lewis, Fountain, & Dana, 1999). Melt is reportedly greatest from the faces of ice cliffs comparative to glaciers’ surfaces during early and late summer and cool midsummer periods because they receive more intense solar radiation (Fountain et al., 1999).

Melt from ice cliffs varies spatially and temporally. The length and orientation of terminus ice cliffs is significant with regards to meltwater generation (Chinn, 1981). Solar aspect and topographic shading influence local diurnal melt cycles (McConchie & Hawke, 2002; Gooseff et al., 2011). Meltwater pulses occur when the sun shines directly on the cliff face.

With regards to sublimation, it accounts for a significant proportion of ablation from glaciers’ surfaces, but ablation from ice cliffs is almost entirely due to melt (Lewis, Fountain, & Dana, 1999). Sublimation from ice cliffs is hampered by the reduced magnitude of surface winds and increased specific air humidity, comparative to glaciers’ surfaces (Lewis, Fountain, & Dana, 1999).

6. CRYOCONITE HOLES Cryconite holes form when sediment accumulated on a ice-covered surface attracts solar radiation and causes the surrounding ice to melt, creating water-filled depressions (Fountain et al., 1999; Howard-Williams & Hawes, 2007). They form in the ablation zones of glaciers with dirty surfaces and on ice shelves, where shortwave radiation is the dominant source of energy. Such environments occur within the Ross Sea Region of Antarctica, and in other polar and alpine areas (MacDonell, Fitzsimons, & Mölg, 2013; Vincent and James, 1996).

The differentiation between cryconite holes and ponds is uncertain in literature. Howard-Williams and Hawes (2007) consider that cryconite holes are 1cm to 1m in diameter and as deep as 0.6m, but other literature does not differentiate between the two. Essentially, when cryconite holes merge they form supgraglacial ponds.

The dominance of shortwave radiation in the Ross Sea Region means that the energy budget is sensitive to small changes of surface albedo caused by sediment accumulation on the surfaces of glaciers, even in low volumes (MacDonell, Fitzsimons, & Mölg, 2013). Sediment accumulation causes a more efficient conversion of shortwave radiation to sensible heat, and associated melt.

It follows that sediment accumulation on glaciers has a significant impact on the hydrological system. The presence of sediment of the surfaces of glaciers has been found to influence both earlier melt and a longer melt season, causing ablation to be nine times greater than for clean ice surfaces (MacDonell, Fitzsimons, & Mölg, 2013). Little information is available on the spatial variability of cryconite holes in the Ross Sea Region, but they certainly occur on the Wright Lower Glacier in Wright Valley (MacDonell, Fitzsimons, & Mölg, 2013). Little sediment is reportedly present on glaciers in Taylor Valley (Fountain et al., 2006), which may mean cryconite holes are less frequent.

Sediment accumulation is highest in winter in most parts of the Ross Sea Region, under the influence of more intense katabatic wind events (Fountain et al., 1999; MacDonell, Fitzsimons, & Mölg, 2013). Cryconite holes develop at the onset of melt in the austral summer. If surface melt occurs followed by refreezing, as is common in early summer, sediment is trapped within the ice, resulting in subsurface melting and more rapid development of cryconite holes (MacDonell, Fitzsimons, & Mölg, 2013). The sediment is flushed with meltwater flows during the austral summer.

The cryconite holes can form systems, whereby they act as liquid water storage reservoirs and conduits, draining the glacier surface. As melt progresses throughout the summer, sediment is flushed away and the volume of sediment in the ablation zone reduces. Therefore, sediment has lesser impact on melt processes as the austral summer progresses, than at the start of the melt season (MacDonell, Fitzsimons, & Mölg, 2013).

7. MELTWATER STREAMS Significant melting is reported to occur below 1500m, but is unusual above this elevation (Chinn, 1981). Below 1500m, channelled streams drain glaciers and flow into lakes, providing a critical hydrological linkage (Fountain et al., 1999; Gooseff, et al., 2011).

At least 24 ephemeral streams are contained within Taylor Valley alone (Fountain et al., 1999), although the total number of streams within the McMurdo Dry Valleys is unknown. Most of the streams in Taylor Valley are reportedly first order, meaning extensive networks have not formed (Conzetto et al., 2006). The maximum temperatures of streams in Lake Fryxell Basin in Taylor Valley reportedly range from 8 to 15°, comparable to glacial and Arctic streams (Cozzetto, Bencala, & Gooseff et al., 2013).

Melt only occurs for approximately 8-12 weeks per year between November and February, but mainly in December and January, influencing ephemeral streams. Most streams show daily signals, with peak discharge in the afternoon (McConchie & Hawke, 2002; Howard-Williams & Hawes, 2007). Streams can also freeze on a daily basis (Howard-Williams & Hawes, 2007).

Gooseff et al. (2011) describe streams of the McMurdo Dry Valleys, and Fountain et al. (1999) provides a detailed description of streams in Taylor Valley. Both of these descriptions are set out below:

Streams of the McMurdo Dry Valleys: “…stable and well-defined, ranging in length from <1 km to over 2 km… Many dry valley streams have extended reaches with wide channels where the streambed is armo[u]red by a stone pavement … and most of the sediment moved in the channels comes from undercutting of the banks. These very stable sections of stream are often covered by microbial mats…” (Gooseff et al., 2011).

Streams of Taylor Valley (within the McMurdo Dry Valleys) from a glacier to a lake:  “At the base and sides of the source glaciers, the streams flow along the moraine and through or around the calved ice. These streams are often frozen over with a thin ice cover.  In areas of ice-bound moraine, there is no alluvium, and the stream flows over and around the frozen rocks.  In steep gradient reaches, the active channel is approximately 5–20 m across, with steep stream banks at the angle of repose of the alluvium. Large jumbled rocks are present in the streambed, with deposited sediment abundant at the margins of the active channel.  In moderate-gradient reaches, the active channel is composed of rocks that are wedged together in a flat stone pavement, with steep stream banks at the angle of repose of the alluvium and less sediment deposition than in steep gradient

reaches.  In both steep and moderate gradient reaches, the streams can cut through a perched delta containing organic-rich sediment.  In shallow-gradient reaches near the lakes or second-order streams in valley bottoms, which receive sediment from tributaries, a sandy braided channel exists, with low banks at the angle of repose of the alluvium.” (Fountain et al., 1999).

The description by Gooseff et al. (2011) generally aligns with the moderate-gradient reaches described by Fountain et al. (1999). The stone pavements in these reaches form in response to freeze-thaw cycles working alluvium – “larger rocks are rotated until the larger sides are upward and these rocks become wedged together across the streambed” (Fountain et al., 1999, pp. 965).

Topography is significant because the stone pavements do not form in steeper reaches, which is likely the result of the hyporheic zone draining more readily and high flow velocities during peak flows eroding sand and destabilising the rocks (Fountain et al., 1999). Topography is often asymmetrical within basins (Fountain et al., 1999; Levy et al., 2014). The south side of basins are north-facing and receive increased solar radiation, influencing freeze-thaw processes and associated gentler slopes, stone pavements, cyanobacterial mats and algal biomass (Fountain et al., 1999). The north side of basins are colder and pole/south-facing, influencing mass wasting processes and associated steeper slopes. For example, Levy et al. (2014, pp. 155) describe: “steep slopes (20-25°) and exposed bedrock, with thin, colluvial soil cover resulting from debris-flows and debris avalanches” on the north side (south-facing slope) of Lake Hoare Basin, Taylor Valley. Freeze-thaw processes influence a gentler slope (5- 10°) and thicker soils on the south side (north-facing slope) of the Lake Hoare Basin (Levy et al., 2014).

In addition to topography, elevation and basin size are also important. In the Taylor Valley, smaller glaciers occur on the north-facing Kukri Hills, comparative to the Asgard Range. The Asgard Range does not have the same north-facing aspect and is also at a higher elevation with larger accumulation basins, influencing larger glaciers (Fountain et al., 1999).

It is noted that although mass wasting processes, including debris flows, have been reported in literature (Dowdeswell et al., 2008; Levy et al., 2014), there is virtually no description of channel confinement, which is critical to debris flow processes.

8. THE ONYX RIVER The Onyx River is the largest river in the region with regards to discharge, despite discharge being very low – with maximum discharge less than 20m3s-1 (Chinn, 1981; Mosley, 1988). Chinn (1981) provides a description of the Onyx River as follows:

“The river arises in the Wilson Piedmont-Wright Lower Glacier area and passes through a proglacial lake [Lake Brownworth] to a flow recorder 1 km downstream of the lake… It continues inland for some 28 km to a flow- 18

measuring weir sited close the point of discharge into Lake Vanda [between Lake Bull and Lake Vanda]. Minor streams from alpine glaciers along the south side of the Wright Valley contribute up to an estimated 10% of the Onyx River flows.”

The Onyx River’s morphology is variable along its length, but it is commonly described as ‘braided’ as per Chinn (1981), which reflects its morphology between Bartley Glacier and Lake Bull (Mosley, 1988). The braided morphology generally aligns with the shallow-gradient reaches described by Fountain et al. (1999) (refer to section 7 above).

Along much of the River’s length, it flows along a sandur, with sand moving down the centre of the channel over gravel armour. Sand principally moves in dunes, but also in ripples or sheets, and active lateral erosion, bar construction and bed scouring (Mosley, 1988). Similar processes have been described for streams in Miers Valley, with ripples, sand and dunes causing bedload flux under higher discharges (Hawke & McConchie, 2001). In the Onyx River, the most common dune extends across the width of mobile sand with an almost plane bed face and a sharp downstream face (45° to the flow), albeit bedform type being difficult to classify (Mosley, 1988). Inactive sandur surfaces have been identified from ice-wedge contraction polygons and former channels, although a channel may be active over multiple seasons (Mosley, 1988).

Similar to meltwater streams, the Onyx River is ephemeral, flowing for up to 90 days per year. The seasonal flow regime is relatively consistent over this period, rising steadily to a peak at the end of December and falling steadily to the middle of February when flows cease (Chinn, 1981; Mosley, 1988). Despite discharge being relatively low, fluvial activity and associated low-intensity erosive processes have modified the landscape over the course of approximately 0.5 million years, since the McMurdo Dry Valleys have been ice-free. Outwash plains and fans in Victoria Land are evidence of such processes (Rains, Selby, & Smith, 1980). Diurnal and other temporal variations occur, in addition to the seasonal flow signature. Diurnal fluctuations are affected by solar radiation, and fluxes outside the diurnal cycle are associated with weather conditions. Formation and breaking of small ice-dams exert a non-climatic influence on discharge, with abrupt hydrograph peaks associated with ice-dams breaking and releasing flows downstream (Mosley, 1988).

Deficits between upstream and downstream discharge occur (Gooseff et al., 2011), and sediment transport is highly variable in space and time, even under constant flow conditions (Mosley, 1988). The deficit between upstream and downstream discharge is up to 1,700,000m3 per day (Gooseff et al., 2011). This reflects storage in hyporheic zones, and processes of evaporation and sublimation discussed with regards to the hydrological system overview above (Mosley, 1988). In addition, heavy early season snowfalls may prevent meltwater from reaching lakes, as is reported to have occurred in 1977-78 for Lake Vanda (Chinn, 1981).

Variability in sediment transport rates under constant flow conditions indicates that other hydraulic conditions, independent of flow, are significant. Sediment supply is thought to be a significant factor affecting transport rates in the Onyx River, especially supply of sand (Mosley, 1988, pp. 56). The is consistent with an investigation of

bedload transport in the meltwater streams of Miers Valley, which also found that sediment supply was a limiting factor (Hawke & McConchie, 2001). Sediment yield is estimated to be over two orders of magnitude less than for Arctic and Alpine Proglacial Rivers. It is crudely estimated to be 3400t/y-1 (Mosley, 1988).

Changes in slope and sedimentary deposits, such as moraines, alluvial fans and deltas, both are a consequence of, and exert an effect on, sediment supply. Mosley (1988) found principal sediment sources were “lateral migration and caving banks, tributary streams, and windblown sand”, with erosion of the channel pavement being virtually insignificant. In Miers Valley, Hawke & McConchie (2001, pp.1) also found that undercutting and collapse of caving banks was significant under elevated discharges, resulting in “random pulses of sediment”. This suggests that similar processes may occur throughout the Dry Valleys of the Ross Sea Region.

9. LAKES 9.1 DEFINING LAKES The McMurdo Dry Valleys contain several lakes up to 75m deep, most of which are closed-basins, formed in depressions or behind glacial dams (Chinn, 1981; Gooseff et al., 2011; Vincent and James, 1996). The geo-data available from the Antarctic Digital Database on the Lakes is limited; therefore a detailed map has not been produced as part of this investigation. However, an upcoming project Assessing Sensitivity to Change in the Dry Valleys will improve the geo-data in the near future.

The Lakes have 3-6m of permanent ice cover and range from saline to brine to freshwater in composition (Fountain et al., 1999; Green and Lyons, 2009). Salinity enables lakes to remain ice-free when air temperatures are below freezing, such as Deep Lake in the Vestfold Hills which is reportedly ice-free at -18°C (Howard-Williams & Hawes, 2007). During winter lakes are reportedly the only liquid water sources in the McMurdo Dry Valleys, and this is enclosed by ice.

Waterhouse (2001) differentiates lakes from ponds in the Ross Sea Region, principally on the basis that lakes retain liquid water over winter and ponds do not. However, Waterhouse (2001) recognises that dry bottomed lakes, which freeze completely over winter, are outside the scope of this classification system. To address this, Waterhouse (2001, pp. 4.61) proposes that dry bottomed lakes are classified on the basis that: “they occupy a clearly defined basin, are fed by meltstreams, and are too large to be considered as ponds.”

9.2 TAYLOR VALLEY LAKES In Taylor Valley, closed-basin Lakes Fryxell, Hoare, and Bonney are remnant of the massive glacial Lake Washburn, which existed during the LGM (Kellogg et al., 1980). Debris deposited by the Ross Ice Sheet formed a low ridge along the marine outlet of Taylor Valley, creating the enclosed basin that is occupied by Lake Fryxell (Fountain et al., 1999). Lake Fryxell, at the eastern end of Taylor Valley, is separated from Lake Bonney, at the western end, by a 700m long bedrock ridge – the Nussbaum Riegel (Fountain et al., 1999; Fountain, Nylen, & MacClune et al., 2006). Lake Hoare is

smaller than Lakes Fryxell and Bonney, and is ice-dammed. Canada Glacier forms part of the ice-dam and if it retreated, Fountain et al. (1999) consider that Lake Hoare would flow east into Lake Fryxell. The lakes all demonstrate a net positive energy balance and are currently rising (Gooseff et al., 2011).

The salinity of the lakes are influenced by their drawdown approximately 1200 years ago when they existed as small hypersaline lakes (Fountain et al., 1999). Lake Bonney is hypersaline at depth, Lake Fryxell is brackish at depth, and Lake Hoare is essentially a freshwater system (Green and Lyons, 2009). Spatial variability in water temperature is high, ranging from 0-1° in Lake Hoare to above 23° in the bottom waters of Lake Vanda, owing to solar heating (Vincent and James, 1996).

9.3 INPUTS AND OUTPUTS Meltwater streams and subglacial melt from wet-based glaciers feed lakes, and streams also act as vectors of heat, generating a thermal wave each day flow reaches the connecting lake (Conzetto et al., 2006). In closed basins, the only outlets are via sublimation or evaporation. Lake levels are a function of the balance between precipitation and melt inputs, and sublimation and evaporation outputs, and are sensitive to small changes in inflow (Fountain et al., 1999). Melt occurs in the austral summer under the influence of warmer temperatures and 24-hour solar radiation, driving moat formation (Gooseff et al., 2011). Sublimation is a function of wind, temperature, radiation, and humidity, and is significant in winter (Gooseff et al., 2011). Evaporation occurs from free water in moats or perched meltwater.

Following the onset of summer melt, 1-10m wide moats form around the periphery of lakes (Fountain et al., 1999; Gooseff et al., 2011; Vincent and James, 1996), and up to 40% of the ice-cover can liquefy (Priscu, Fritsen, & Adams et al., 1998). At the end of the summer season the moats re-freeze following the maximum summer lake level, around the time inflows from meltwater streams cease. However, if the ice cover is sufficiently thick, meltwater perches on top of the lake and a moat does not form. This occurs on Lake Vida in the Victoria Valley which has greater than 15m of ice-cover (Gooseff et al., 2011).

A lake ice coveyor mechanism can affect sediment deposition in the proglacial lakes, whereby ice cover rafts glacial drift across the lake towards the centre and distal edges. This has been reported in the Garwood Valley, resulting in laminated fine grained till filtering through the ice, overlaid by lacustrine sediments (Levy et al., 2013). Glaciological or hydrological drivers can trigger the lake ice coveyor mechanism, such as glacier advancement or lake levels rising (Levy et al., 2013).

9.4 DRIVERS OF LAKE LEVEL CHANGE Net lake level rise is currently occurring, and is largely a function of positive summer energy balance, influenced by air temperatures (Gooseff et al., 2011). However, there are seasonal variations, with summer rise balancing out lake level drops during the winter and shoulder seasons. The shoulder seasons are associated with the most rapid lake level drops because greater solar radiation and warmer temperatures, comparative to winter, increase evaporation.

Lakes respond to minor changes in heat balance, meaning that lake level change is used towards understanding local cycles of warming and cooling since the massive ice retreat and associated lake level falls coming out of the LGM (Chinn, 1981; Howard- Williams & Hawes, 2007). Perched deltas provide sedimentary records of paleo- climate conditions when streams deposited sediment loads into lakes with higher water levels than observed today (Fountain et al., 1999; Kellogg et al., 1980; Levy et al., 2013). Lake level change has been correlated with temperature records from the Taylor Dome ice core (Howard-Williams & Hawes, 2007).

Sharp cooling events are reported at 9ka and 6.5ka (Howard-Williams & Hawes, 2007). Warming is reported to have triggered the commencement of a sustained period of lake level rise around 5ka to 6ka, resulting in maximum lake level depths around 3ka (Fountain et al., 1999). This was followed by a low around 1-1.5ka when colder, drier conditions prevailed. Subsequent rising continues today, under the influence of approximately 2°C of temperature warming (Chinn, 1981; Fountain et al., 1999; Howard-Williams & Hawes, 2007). Lake Vanda is an exception to this regional trend, reported to have reached water highs around 5.6ka and between 2.9ka and 2.1ka, with significant lowering in the interim (Chinn, 1981).

In addition to climatic variations, lakes also respond to displacement by glaciers, rising as glaciers enter enclosed systems, and to subglacial flow from wet-based glaciers (Chinn, 1981). The largest lake level changes are reportedly associated with glacial displacement. Such displacement could occur in Lake Bonney where the Taylor Glacier terminates, or Lake Hoare where the western lobe of the Canada Glacier terminates for example (Fountain et al., 2006). Subglacial flow from wet-based glaciers can result in year-round replenishment of the lakes, overlaid on the climate signal described above.

10. PONDS Ponds are reportedly “the most common type of water body in inland Antarctica” (Howard-Williams & Hawes, 2007, pp. 209). A variety of ponds occur within the Ross Sea Region, both supra-glacially and in ice-free areas, from high altitudes to the coast (Broady, 1989; Howard-Williams & Hawes, 2007; Vincent and James, 1996). Some ponds are permanently ice-covered like the lakes, but others are open-water in the austral summer (Howard-Williams & Hawes, 2007).

Significant systems of ponds have been reported in the ablation zone of the McMurdo Ice Shelf (Jungblut, Allen, & Burns et al., 2009; Hawes et al., 2011; Vincent and Howard-Williams, 1994); in the Pyramid Trough Region at the southern end of the McMurdo Dry Valleys some 75km from the open sea (Vincent and Howard-Williams, 1994); in Garwood Valley – one of the southern McMurdo Dry Valleys (Levy et al., 2013); in the Darwin Glacier Region some 300km south of McMurdo Sound (Vincent and Howard-Williams, 1994); and in the ice-free regions of Cape Bird, Cape Royds, and Cape Crozier on Ross Island (Broady, 1989) (Figure 2).

The existence of ponds on the McMurdo Ice Shelf is not surprising, given it is an ablation zone covered in sediment; the cryconite hole systems described above likely

lead to the development of seasonal ponds. Hawes et al. (2011) found that the ponds on the McMurdo Ice Shelf refreeze relatively rapidly following the commencement of ice formation, with only 0-15% of the volume of each pond still existing in liquid form after two months of ice formation. Ice formation was found to be reasonably consistent between ponds, and variability in the volume of liquid water was found to primarily be a function of depth (Hawes et al., 2011).

The Pyramid Trough and Darwin Glacier Regions are described by Vincent and Howard-Williams (1994). In a deep valley at 400m altitude, the Koettlitz Glacier feeds the Alph River which flows into proglacial Trough Lake in the Pyramid Trough Region. A more extensive description of ponds in the Darwin Glacier Region is provided as follows: “On visiting this area we found pools … of clear water up to 0.5 km long and 0.1 km wide mostly overlain by a thin (<0.5m) layer of ice. Several streams up to 1 m wide of greenish coloured water were flowing over the glacier surface. Ponds with open water up to 0.5 km long were also observed in the pressure ridges further up the Darwin Glacier at the confluence with the Hatherton Glacier. Diamond Hill is an ice-free region with many small valleys, each of which contains small ponds and interconnecting streams… A large, permanently ice- covered lake, Lake Wilson, lies at the base of this region between Diamond Hill and the Ross Ice Shelf. A deep moraine-filled valley was visited to the south of Mount Ash in the Darwin Mountains, at an altitude of 1000 m. The highly undulating surface of this valley was occupied by several hundred ponds of variable size, but typically 40 m by 40 m or less.”

Ponds in the ice-free regions of Cape Bird, Cape Royds, and Cape Crozier on Ross Island have been divided into four classes with regards to their catchments by Broady (1989, pp. 80-83), quoted below. The first category is consistent with descriptions of “closed-basin ponds, lying in shallow depressions” by Hawes et al. (2011, pp.235).

1. Typical Ponds “Most numerous were ponds which have been termed ‘typical ponds’. These were found in depressions in glacial till…, volcanic tephra and lava, and beach sands. They numbered 45, 145 and 45 at C. Bird, Royds and Crozier respectively. They include all ponds other than those in the three remaining categories described below. Several were visited by penguins and in particular by skuas and received some nutrient enrichment from these sources. Their areas varied widely from 1 to 4000 m2. All were shallow with a maximum depth of < 2 m. Some dried out during the study period. Ponds in areas affected by sea-spray were predominantly of high conductivity (300 to 2000 m Sm-1) and of saline taste, whereas outside these areas the majority were of lower conductivity (50-300 m Sm-1) and of fresh taste…”

2. Small Ponds “Small ponds formed along the shoreline due to the melting of sea-ice which had become grounded on sandy beaches… These ‘shoreline ponds’ were frequent in the northern half of C. Bird, with about 25 in total, infrequent at C. Royds, 5 in total, and none were observed at C. Crozier. Strongwinds blow a

layer of dark sand over the grounded ice. Solar heating of this causes the underlying ice to melt, forming ponds of 0.2-200 m2 and up to 30 cm deep… On complete melting of the ice the water drains away and only a smaIl depression remains as an indication of the previous presence of a pond. These ponds were generally brackish to saline.”

3. Enriched Ponds “Within and directly adjacent to the nesting area of penguins…there were turbid, highly nutrient and organically enriched ponds... The catchments of these ‘rookery ponds’, and the ponds themselves, contained substantial guano deposits and scattered bird carcasses. They numbered 9, 1 and 4 at C. Bird, Royds and Crozier respectively. All were saline with conductivities >300m Sm- 1.”

4. Cryconite Ponds “Mineral sediments (‘cryoconite’) melting into glacier ice form ‘cryoconite ponds’… These were found only in the ice bordering the north-eastern edge of C. Bird. Here the hard ice surface was darkened by substantial mineral deposits which had melted into the ice on flat areas (Fig. 14). The ponds were generally cylindrical, 20-40 cm diameter and 20-30 cm deep, although occasionally they attained 5 m2 in surface area. The waters were of low conductivity, less than 50 m Sm-1. Their absence at C. Royds and C. Crozier was probably due to the sloping topography of most ice-fields and the relatively clean ice surfaces.”

The pond classification system developed by Broady (1989) for the purposes of investigating aquatic and terrestrial vegetation is partially based on morphological process. For example, small ponds and cryconite ponds have been categorized largely based on the way they form by melt of sea ice or melt of sediments into a glacier respectively. However, enriched ponds appear to have been categorized based on ecological processes and associated nutrient enrichment, with all other ponds being lumped into the first category of ponds that occur in depressions.

11. SUBSURFACE FLOWS Sub-surface flow is limited as the landscape is underlain by permafrost to depths greater than 100m, and up to at least 600m. Permafrost may also constrain geomorphic effectiveness and associated channel development as has been found in the Arctic (Crawford and Stanley, 2014). Permafrost melt occurs in the austral summer, resulting in approximately 10-70cm of unfrozen soils - commonly referred to as the active layer (Gooseff et al., 2011; Levy et al., 2014). Dry permafrost occurs in the McMurdo Dry Valleys, meaning it has very little water content, and ice cemented permafrost also occurs which gives rise to soil moisture upon melt (Levy et al., 2014).

As described with regards to the hydrological system above, hyporheic zones are thawed areas adjacent to, and thawed by, stream channels. They act like a porous sponge, soaking up water before surface flows eventuate when saturation is reached, and maintain cooler water temperatures than surface streams (Cozzetto et al., 2006).

Water tracks occur within the hyporheic zone, routing shallow groundwater downslope, influencing sediment and nutrient flux. They may be defined as follows: “Water tracks are topographically-controlled permafrost groundwater conduits that route water derived from snowmelt, ground ice melt, and brine migration downslope, above the ice table, in polar environments (Levy et al., 2014).

Despite limitations associated with permafrost, subsurface water tracks have been found to provide an important hydrological connection between upland terrains and valley bottoms in Taylor Valley, moving snowmelt and ground ice melt downslope (Levy et al., 2014). They have been found to provide long-range spatial structure to terrestrial hillslope ecosystems, but elevated soil salinity within these zones limits faunal biological activity (Levy et al., 2014).

12. A PILOT FRAMEWORK FOR DEVELOPING A CLASSIFICATION SYSTEM As indicated in the introductory section, fluvial classifications can be form-based or process-based (McDavitt, 2004). It has been shown in the preceding sections that qualitative form-based classifications have been inadvertently initiated in literature relating to the Ross Sea Region, predominantly to assist with ecological applications (e.g. Broady, 1989; Fountain et al., 1999; Waterhouse, 2001). However, quantitative and/or process-based classifications have not been attempted. Process-based classifications allow predictions of preferential response to be made (Montgomery and Buffington, 1997). In the Ross Sea Region, this would assist with addressing the problems of assessing and forecasting spatial variability of fluvial response to climate forcing and human impact.

In order to develop a process-based classification system for the Ross Sea Region, knowledge of the hydrological regime and fluvial morphology need to be integrated and the dominant forces driving morphology understood (refer to the following sub- sections). Once developed, the classification system would enable an understanding of dominant processes driving fluvial morphology in reaches or water bodies’ throughout the Ross Sea Region to be gained by visual observation. Understanding the dominant processes would, in turn, assist with forecasting how fluvial systems in the Ross Sea Region may preferentially respond to climate change and human impact.

12.1 STREAMS AND RIVERS Transport- and Supply-limited Conditions Unlike other environments, morphological response in the Ross Sea Region of Antarctica is severely transport-limited as discharge only occurs for 8-12 weeks during austral summer. Based on research undertaken in non-polar environments (Montgomery and Buffington, 1997), it may be hypothesized that transport-limited first-order streams in the Ross Sea Region have morphologies that are heavily affected by hill-slope processes.

Although the temporal variability of the hydrological regime for the Ross Sea Region has been investigated, very limited information exists that describes the spatial

distribution of channel morphologies and/ or investigates hill-slope processes, making coupling of these into a workable classification system impossible with the current state of knowledge. Debris flows and debris avalanches are reported to occur in the Ross Sea Region (Levy et al., 2014), and the description of steep gradient reaches provided by Fountain et al. (1999) appears consistent with such events. However, geomorphological investigations (such as sedimentary analyses) to determine the significance, origin and timing of mass wasting events, and work to understand how such events unfold in polar environments (where glacially worked sediment may available, but the ground is almost completely frozen) are severely lacking, let alone investigations correlating these events with fluvial morphologies.

Given significant transport limitations, surplus sediment supply may be expected. However, Mosley (1988) and Hawke and McConchie (2001) have demonstrated that this is not necessarily the case, because sediment transport rates vary under constant flow conditions. Moreover, sediment yield in the Antarctic has been found to be approximately two orders of magnitude less than in the Arctic. Streams and rivers in the Ross Sea Region may therefore be considered both transport- and supply- limited. Whether transport-limited conditions coupled with supply-limited conditions in Antarctica result in different morphologies, comparative to when only one factor is limited, is a hypothesis worth testing to improve knowledge of morphological response.

Drivers of Spatial Organisation In some ways, the Antarctic hydrological system may be seen as simplified as it is not subject to forcing by vegetation or woody debris, and is only subject to very limited anthropogenic disturbance. However, as shown in the preceding sections, the hydrological regime and resulting fluvial morphologies of the Ross Sea Region are complex for other reasons.

Of relevance to the hydrological regime, at a fundamental level, is the presence or absence of liquid water, which provides means for channels, ponds and lakes to develop. Associated with this are climatic and physiographic factors influencing the volume and speed of melt (given very limited precipitation is insignificant to flow). In other words, the hydrological regime in the Ross Sea Region is spatially complicated by local climatic conditions and physiographic factors.

Energy balance drives discharge, and temperature and wind are significant controlling factors of the hydrological regime. Precipitation is very low and, unlike other environments, can actually limit discharge by reducing the albedo of glaciers’ surfaces and thus hampering melt. With regards to physiographic factors, elevation, topography, aspect, catchment size, ice-cover and ice cliffs have all been shown to be significant to the spatial distribution of melt above. The extent of the sponge-like hyporheic zone, which is associated with channel length, effects whether melt processes produce surface flows. However, whether spatial variability of melt, and the proportion of this melt that eventuates as surface flows, drive fluvial morphologies is less certain.

Evidence gathered in this report suggests that fluvial morphology does respond to drivers other than flow. For example, literature demonstrates that freeze-thaw cycles

produce stone pavements. As described above, literature also demonstrates that sediment supply may be a relevant factor. Channel descriptions by Fountain et al. (1999) suggest that there may be a down-valley arrangement of channel reaches, affected by slope, similar to other environments. This is not surprising given slope is a parameter of stream power (the ability to erode and transport material), as is discharge.5

The other widely reported channel form is braiding. Drivers of braided river morphologies observed in the Ross Sea Region can likely be deduced from research in other environments. Smith (2000) identified that research should move towards understanding how braided rivers respond to various environmental forcing. Synthesising research outside of the Antarctic context since this time will likely assist with understanding how the braided rivers of the Ross Sea Region may preferentially respond to environment change, and identify research areas that need to be addressed.

Discharge events that are not directly “climatic” are reported to occur in the Ross Sea Region, including: when a glacier flows into and displaces a lake, jökulaups (ice-dam floods), and basal meltwater drainage of wet-based glaciers. Whilst these elevated flow events are reported in literature, it is unknown to what extent they influence the fluvial geomorphology of the Ross Sea Region. Responses to elevated flows could be transient, or perhaps more likely given transport-capacity is usually severely limited, could trigger evolutionary change. The latter should be tested as a hypothesis in order to assist with developing a process-based classification system.

12.2 PONDS AND CRYCONITE HOLES There is some confusion in literature as to what constitutes a pond and what constitutes a cryconite hole. If the pond is formed as a result of the cryconite hole expanding, making an arbitrary size differentiation between a “hole” and a “pond” is of limited use in determining how the hydrological regime may respond. Of more significance, is determining the proportion of ice-cover in the Ross Sea Region affected by cryconite holes / ponds and forecasting whether and where this will increase or reduce with regards to climate forcing and changes in sediment cover. For example, elevated sediment cover associated with glacial processes, hill-slope processes, wind distribution, and/or hydrological processes would likely elevate cryconite hole and associated pond formation. This may, in turn, lead to more rapid environmental change as ice melts and biota are able to survive in more expansive areas of liquid water.

Ponds in ice-free areas have been differentiated from lakes in literature on the basis of whether they seasonally freeze to the bottom, and if they do, on the basis of whether their size, basin occupancy, and meltwater stream inputs are insufficient for them to be classified as lakes (Waterhouse, 2001). Whilst this is of limited relevance

5 Ω = pgQS Ω is stream power, ρ is the density of water (1000 kg/m3), g is acceleration due to gravity (9.8 m/s2), Q is discharge (m3/s), and S is channel slope.

to geomorphology, whether water bodies freeze to the bottom is significant for ecologically focused environment change research. This is because lakes are reportedly the only source of liquid water in winter in the Ross Sea Region, providing habitat for aquatic microbial life. Differentiating ponds and lakes with regards to a size variable that has not even been defined, however, is not at all useful.

The other categories identified by Waterhouse (2001), relating to basin occupancy and meltwater stream inputs, are more relevant to fluvial geomorphology. Occupancy of a basin indicates that the water body is more than transient, and is likely to remain intact, albeit fluctuations of volume, potentially over several thousand years. The occurrence of meltwater streams indicate that the water body will be replenished, which further allows it to remain a permanent hydrological feature of the Region. It is, therefore, suggested that ponds in ice-free areas may be distinguished from lakes on grounds of whether they are likely permanent over hundreds to several thousand or more years, or transient features of the environment, lasting for only a season or a few years. Further research into the formation and desiccation of ponds, and their associated impact on the land’s surface, is required in order to better understand their role in both the hydrological function and geomorphology of the Ross Sea Region. Such investigations are justified as the abundance of ponds reported in literature suggests that they are a significant part of the system.

12.3 LAKES Processes forming lakes are better understood, and can essentially be divided into two categories: lakes enclosed in topographical depressions, and ice-dammed lakes. Distinguishing between these two types of lakes is significant with regards forecasting morphological response to environmental change. Whilst lakes formed in topographical depressions rise and fall in response to climate forcing, leaving records of climate change in perched deltas, ice-dammed lakes can drain rapidly on collapse of the dam. The morphological response of the latter is poorly understood, although such events are widely reported to occur in the Ross Sea Region. In order to develop a comprehensive process-based classification system, further research into the impact of ice-dam floods needs to be undertaken.

13. CONCLUSION This report has presented an overview of the hydrological regime for the Ross Sea Region, coupled with information on the Region’s fluvial geomorphology. On the basis of this information, a pilot framework for the development of an integrated process- based classification system has been developed. The pilot framework has identified several gaps in knowledge that need to be addressed through both qualitative and quantitative investigations before the framework can be developed further and a process-based classification system established. In no particular order, these gaps include knowledge of: the spatial distribution of channel morphologies in the Ross Sea Region; fluvial morphological behaviour under heavily transport- and supply-limited conditions; the significance, timing and origin of hill-slope processes; the formation and desiccation of ponds, and their associated impact on the land’s surface; whether the spatial variability of melt, and the proportion of this melt that eventuates as surface flows, drive fluvial morphologies, or whether other processes exert a greater 28

control; and associated with the latter whether events that are not directly climate/melt-driven, including when a glacier flows into and displaces a lake, jökulaups (ice-dam floods), and basal meltwater drainage of wet-based glaciers, have a transient or evolutionary effect on fluvial morphology in the Ross Sea Region.

14. REFERENCES

Addy, S., Soulsby, C., & Hartley, A. (2014). Controls on the distribution of channel reach morphology in selectively glaciated catchments. Geomorphology, 211, 121-133. Anderson, J. B., Conway, H., Bart, P. J., Witus, A. E., Greenwood, S. L., McKay, R. M., . . . Jakobsson, M. (2014). Ross Sea paleo-ice sheet drainage and deglacial history during and since the LGM. Quaternary Science Reviews, 100, 31-54. Araneo, D. C., & Compagnucci, R. H. (2008). Atmospheric circulation features associated to Argentinean Andean rivers discharge variability. Geophysical Research Letters, 35(1). Bockheim, J. G. (2014). Distribution, properties and origin of viscous-flow features in the McMurdo Dry Valleys, Antarctica. Geomorphology, 204, 114-122. Broady, P. A. (1989). Broadscale patterns in the distribution of aquatic and terrestrial vegetation at three ice-free regions on Ross Island, Antarctica High Latitude Limnology (pp. 77-95): Springer. Campbell, I., & Claridge, G. (1988). Landscape evolution in Antarctica. Earth-Science Reviews, 25(5), 345-353. Carter, S. P., Blankenship, D. D., Peters, M. E., Young, D. A., Holt, J. W., & Morse, D. L. (2007). Radar‐based subglacial lake classification in Antarctica. Geochemistry, Geophysics, Geosystems, 8(3), pp. 1-20. Chinn, T. (1981). Hydrology and climate in the Ross Sea area. Journal of the Royal Society of New Zealand, 11(4), 373-386. Clark, P. U., Dyke, A. S., Shakun, J. D., Carlson, A. E., Clark, J., Wohlfarth, B., . . . McCabe, A. M. (2009). The Last Glacial Maximum. Science, 325, 710-714. Clarkson, P. (2007). Antarctic: definitions and boundaries. In: B, Riffenburgh (Ed.), Encyclopedia of the Antarctic. (pp. 47-51). New York: Routledge. Colesie, C., Gommeaux, M., Green, T., & Büdel, B. (2014). Biological soil crusts in continental Antarctica: Garwood Valley, southern Victoria Land, and Diamond Hill, Darwin Mountains region. Antarctic Science, 26(02), 115-123. Conovitz, P. A., McKnight, D. M., MacDonald, L. H., Fountain, A. G., & House, H. R. (1998). Hydrologic processes influencing streamflow variation in Fryxell Basin, Antarctica. Wiley Online Library, pp. 93-108. Cooper, A. F., Adam, L. J., Coulter, R. F., Eby, G. N., & McIntosh, W. C. (2007). Geology, geochronology and geochemistry of a basanitic volcano, White Island, Ross Sea, Antarctica. Journal of Volcanology and Geothermal Research, 165(3), 189-216.

Cozzetto, K., McKnight, D., Nylen, T., & Fountain, A. (2006). Experimental investigations into processes controlling stream and hyporheic temperatures, Fryxell Basin, Antarctica. Advances in Water Resources, 29(2), 130-153. Cozzetto, K. D., Bencala, K. E., Gooseff, M. N., & McKnight, D. M. (2013). The influence of stream thermal regimes and preferential flow paths on hyporheic exchange in a glacial meltwater stream. Water Resources Research, 49(9), 5552-5569. Crawford, J. T., & Stanley, E. H. (2014). Distinct Fluvial Patterns of a Headwater Stream Network Underlain by Discontinuous Permafrost. Arctic, Antarctic, and Alpine Research, 46(2), 344-354. Denton, G. H., Bockheim, J. G., Wilson, S. C., & Stuiver, M. (1989). Late Wisconsin and early Holocene glacial history, inner Ross embayment, Antarctica. Quaternary Research, 31(2), 151-182. Doran, P. T., McKay, C. P., Clow, G. D., Dana, G. L., Fountain, A. G., Nylen, T., & Lyons, W. B. (2002). Valley floor climate observations from the McMurdo Dry Valleys, Antarctica, 1986–2000. Journal of Geophysical Research: Atmospheres (1984– 2012), 107(D24),4772, pp. 13-1-13-12. Doran, P. T., Priscu, J. C., Lyons, W. B., Walsh, J. E., Fountain, A. G., McKnight, D. M., . . . Clow, G. D. (2002). Antarctic climate cooling and terrestrial ecosystem response. Nature, 415, 517-520. Dowdeswell, J., Cofaigh, C. Ó., Noormets, R., Larter, R. D., Hillenbrand, C.-D., Benetti, S., . . . Pudsey, C. (2008). A major trough-mouth fan on the continental margin of the Bellingshausen Sea, West Antarctica: the Belgica Fan. Marine Geology, 252(3), 129-140. Fountain, A. G., Dana, G. L., Lewis, K. J., Vaughn, B. H., & Mcknight, D. H. (1998). Glaciers of the McMurdo Dry Valleys, southern Victoria Land, Antarctica: Wiley Online Library. Fountain, A. G., Lyons, W. B., Burkins, M. B., Dana, G. L., Doran, P. T., Lewis, K. J., . . . Priscu, J. C. (1999). Physical controls on the Taylor Valley ecosystem, Antarctica. Bioscience, 49(12), 961-971. Fountain, A. G., Nylen, T. H., MacCLUNE, K. L., & Dana, G. L. (2006). Glacier mass balances (1993–2001), Taylor Valley, McMurdo Dry Valleys, Antarctica. Journal of Glaciology, 52(178), 451-462. Gooseff, M. N., McKnight, D. M., Doran, P., Fountain, A. G., & Lyons, W. B. (2011). Hydrological connectivity of the landscape of the McMurdo Dry Valleys, Antarctica. Geography Compass, 5(9), 666-681. Gore, D. B. (1992). Ice-damming and fluvial erosion in the Vestfold Hills, East Antarctica. Antarctic Science, 4(02), 227-234. Green, W. J., & Lyons, W. B. (2009). The Saline Lakes of the McMurdo Dry Valleys, Antarctica. Aquatic Geochemistry, 15(1-2), 321-348. Hawke, R.M., & McConchie, J.A. (2001). Bedload transport in a meltwater stream, Miers Valley, Antarctica: controls and prediction. Journal of Hydrology (NZ), 40(1), 1-18.

Hawes, I., Howard-Williams, C., & Sorrell, B. (2014). Decadal timescale variability in ecosystem properties in the ponds of the McMurdo Ice Shelf, southern Victoria Land, Antarctica. Antarctic Science, 26(03), 219-230. Hawes, I., Safi, K., Sorrell, B., Webster-Brown, J., & Arscott, D. (2011). Summer–winter transitions in Antarctic ponds I: The physical environment. Antarctic Science, 23(03), 235-242. Hawes, I., Sumner, D., Andersen, D., & Mackey, T. (2011). Legacies of recent environmental change in the benthic communities of Lake Joyce, a perennially ice‐covered Antarctic lake. Geobiology, 9(5), 394-410. Hulbe, C. (2007). Glaciers and ice streams. In: B. Riffenburgh (Ed.), Encyclopedia of the Antarctic. (pp. 463-466). New York: Routledge. Jungblut, A. D., Allen, M. A., Burns, B. P., & Neilan, B. A. (2009). Lipid biomarker analysis of cyanobacteria-dominated microbial mats in meltwater ponds on the McMurdo Ice Shelf, Antarctica. Organic Geochemistry, 40(2), 258-269. Kellerhals, R., Bray, D. I., & Church, M. (1976). Classification and analysis of river processes. Journal of the Hydraulics Division, 102(7), 813-829. Kellerhals, R., Bray, D. I., & Church, M. (1976). Classification and analysis of river processes. Journal of the Hydraulics Division, 102(7), 813-829. Kellogg, D. E., Stuiver, M., Kellogg, T. B., & Denton, G. H. (1980). Non-marine diatoms from late Wisconsin perched deltas in Taylor Valley, Antarctica. Palaeogeography, Palaeoclimatology, Palaeoecology, 30, 157-189. Kiem, A., & Verdon-Kidd, D. (2010). Towards understanding hydroclimatic change in Victoria, Australia–preliminary insights into the" Big Dry". Hydrology and Earth System Sciences, 14(3), 433-445. Knuth, S. L., Tripoli, G. J., Thom, J. E., & Weidner, G. A. (2010). The influence of blowing snow and precipitation on snow depth change across the Ross Ice Shelf and Ross Sea regions of Antarctica. Journal of Applied Meteorology and Climatology, 49(6), 1306-1321. Laizé, C. L., & Hannah, D. M. (2010). Modification of climate–river flow associations by basin properties. Journal of Hydrology, 389(1), 186-204. Lavers, D., Prudhomme, C., & Hannah, D. M. (2010). Large-scale climate, precipitation and British river flows: Identifying hydroclimatological connections and dynamics. Journal of Hydrology, 395(3), 242-255. Leopold, L. B., Wolman, M. G., Wolman, M. G., & Wolman, M. G. (1957). River channel patterns: braided, meandering, and straight. Levy, J. S., Fountain, A. G., Gooseff, M. N., Barrett, J. E., Vantreese, R., Welch, K. A., . . . Wall, D. H. (2014). Water track modification of soil ecosystems in the Lake Hoare basin, Taylor Valley, Antarctica. Antarctic Science, 26(02), 153-162. Levy, J. S., Fountain, A. G., O’Connor, J. E., Welch, K. A., & Lyons, W. B. (2013). Garwood Valley, Antarctica: A new record of Last Glacial Maximum to Holocene glaciofluvial processes in the McMurdo Dry Valleys. Geological Society of America Bulletin, 125(9-10), 1484-1502.

Lewis, K. J., Fountain, A. G., & Dana, G. L. (1999). How important is terminus cliff melt?: A study of the Canada Glacier terminus, Taylor Valley, Antarctica. Global and Planetary Change, 22(1), 105-115. Lyons, W., Tyler, S., Wharton, R., McKnight, D., & Vaughn, B. (1998). A late Holocene desiccation of Lake Hoare and Lake Fryxell, McMurdo Dry Valleys, Antarctica. Antarctic Science, 10(03), 247-256. MacDonell, S. A., Fitzsimons, S. J., & Mölg, T. (2013). Seasonal sediment fluxes forcing supraglacial melting on the wright lower glacier, McMurdo Dry Valleys, Antarctica. Hydrological Processes, 27(22), 3192-3207. McConchie, JA., & Hawke, R.M. (2002). Controls on streamflow generation in a meltwater stream, Miers Valley, Antarctica and implications for the debate on global warming. Journal of Hydrology (NZ), 41(2), 77-103. McDavitt, W. N. (2004). Application of a stream channel classification in a mountainous setting. Utah State University, Department of Geography. McKnight, D. M., Niyogi, D. K., Alger, A. S., Bomblies, A., Conovitz, P. A., & Tate, C. M. (1999). Dry valley streams in Antarctica: ecosystems waiting for water. Bioscience, 49(12), 985-995. Montgomery, D. R., & Buffington, J. M. (1997). Channel-reach morphology in mountain drainage basins. Geological Society of America Bulletin, 109(5), 596- 611. Mosley, M. (1988). Bedload transport and sediment yield in the Onyx River, Antarctica. Earth Surface Processes and Landforms, 13(1), 51-67. Nie, Y., Liu, X., Sun, L., & Emslie, S. D. (2012). Effect of penguin and seal excrement on mercury distribution in sediments from the Ross Sea region, East Antarctica. Science of the Total Environment, 433, 132-140. Nylen, T. H., Fountain, A. G., & Doran, P. T. (2004). Climatology of katabatic winds in the McMurdo dry valleys, southern Victoria Land, Antarctica. Journal of Geophysical Research: Atmospheres (1984–2012), 109(D3). Poveda, G., & Mesa, O. J. (1997). Feedbacks between hydrological processes in tropical South America and large-scale ocean-atmospheric phenomena. Journal of Climate, 10(10), 2690-2702. Prairie, Y., Salm, C. R., Saros, J. E., Fritz, S. C., Osburn, C. L., & Reineke, D. M. (2009). Phytoplankton productivity across prairie saline lakes of the Great Plains (USA): a step toward deciphering patterns through lake classification models. Canadian Journal of Fisheries and Aquatic Sciences, 66(9), 1435-1448. Pritchard, H., Ligtenberg, S., Fricker, H., Vaughan, D., Van den Broeke, M., & Padman, L. (2012). Antarctic ice-sheet loss driven by basal melting of ice shelves. Nature, 484(7395), 502-505. Rains, R., Selby, M., & Smith, C. (1980). Polar desert sandar, Antarctica. New Zealand Journal of Geology and Geophysics, 23(5-6), 595-604. Rosgen, D. L. (1985). A stream classification system. Paper presented at the Riparian ecosystems and their management: reconciling conflicting uses. First North American Riparian Conference, Arizona.

Rosgen, D. L. (1994). A classification of natural rivers. Catena, 22(3), 169-199. Smith, G. S. (2000). Small-scale cyclicity in alpine proglacial fluvial sedimentation. Sedimentary Geology, 132(3), 217-231. Strahler, A. N. (1957). Quantitative analysis of watershed geomorphology. Eos, Transactions American Geophysical Union, 38(6), 913-920. Uttormark, P. D., & Wall, J. P. (1975). Lake classification, a trophic characterization of Wisconsin lakes: National Environmental Research Center. Vincent, W., & James, M. (1996). Biodiversity in extreme aquatic environments: lakes, ponds and streams of the Ross Sea sector, Antarctica. Biodiversity & Conservation, 5(11), 1451-1471. Vincent, W. F., & Howard-Williams, C. (1994). Nitrate-rich inland waters of the Ross Ice Shelf region, Antarctica. Antarctic Science, 6, 339-339. Wagner, T., Bremigan, M. T., Cheruvelil, K. S., Soranno, P. A., Nate, N. A., & Breck, J. E. (2007). A multilevel modeling approach to assessing regional and local landscape features for lake classification and assessment of fish growth rates. Environmental monitoring and assessment, 130(1-3), 437-454. Waterhouse, E. J. (2001). Ross Sea region 2001: A State of the Environment Report for the Ross Sea region of Antarctica. Rangiora, New Zealand: New Zealand Antarctic Institute.